Semantics for Large Banks (1) – Open Banking (video)

This video derived from a recent overview for a large bank with over $400bn in assets. Their new initiative is for Open Banking, and the meeting discussed FIBO Data Model scoping for PSD2 using the FIB-DM, Concept Maps.

The related video for banks with Semantic Centers of Excellence who want to transform their in-house ontologies has been released: Semantics for Extra Large Banks


Transcript

Hello and welcome back to FIB-DM, the Financial Industry Business Data Model. The most popular video on the FIB-DM YouTube channel was Semantics for mid-sized banks. For that reason, I want to continue that series. This installment is called Semantics for Everyone’s Large Bank ELB. Just like the previous video, this is based on a deck I prepared for an actual meeting with a bank. This Bank here is firmly in the EDM Council Tier A bracket with more than four hundred billion dollars in assets, twice the size of the mid-size bank. Now, I found this interesting here because they were in particular interested in Open Banking and how to leverage the FIB-DM for an Open Banking implementation.

One more word about the size brackets: They are a poor proxy for the sophistication of a bank. There are very large banks that do not have Triple-Stores in production. They do not have a Center of Excellence yet and there are midsize financial institutions, in particular in the investment management and hedge funds space, that are very sophisticated. So, I know it’s a poor proxy but kind of popular with the audience. This installment is for the larger banks that have the same problems as midsize banks. There will be a final installment, I will call it extra-large banks XXL banks. This will be for financial institutions who do have Semantics Centers of Excellence; who do have coded in-house transformations from the ontology to data models and vice versa, but who want to compare their transformation to the Configurable Ontology to Data model Transformation. Now back to this presentation, here again, It’s based on a meeting deck and I replaced the bank’s logo and name with a generic, – ELB. Thank you and enjoy the video.

Okay, so why are we here? FIBO and FIB-DM are the industry standard. The Enterprise Data Management Council, the creators of the FIBO, recently endorsed the FIBO Data Model, putting a page for FIB-DM on the EDM Council website. And, as of last week, 664 users from 162 different banks downloaded the Open Source Core version. ELB bank, “Everyone’s Large Bank,” the Data Architecture team downloaded the PowerDesigner model a while ago and you have been testing the Open Source version since.

ELB’s current state challenges and they are typical for many banks. It’s a classic case for an Enterprise Data Model, for reference models. There are thousands of silo database systems with millions of different tables. There’s duplication for Customer and Product reference databases, so you cannot really call them Master systems. The Data Architecture team is trapped in operational support, rather than doing what they want to do, which is providing an integrated design for the bank. They are validating operational schemas rather than integrating them. Those challenges are not just for EAB, but many banks, of smaller and larger size, the lack of management buy-in for Logical Reference Models. Now, all this is what the EDM council, now they do the famous Data management Capability Assessment Model. And it would rate probably a poor data management capability and that impedes competitiveness, innovation, and it is a nightmare for Regulatory Compliance.

The vision is Semantic Enterprise Information Architecture. When we look at Enterprise Architecture, we can break down artifacts by the Use, business design development, by the Type whether the model is conceptual logical physical or whether it’s an implementation artifact. And the level of whether where it applies, Enterprise Department, or project. Now, the Semantic component here is that we place the ontology, the FIFO at the apex. From the ontology we generate consistent RDF schemas for our Triplestores, we generate data models, we generate object models. The new thing here the FIB Concept Map this is a simplified model for our communication with business users. It’s in sync with the data model of the ontology. Now, this use case is about Open Banking. For Open Banking of course we need object models as Open Banking is an API. We need messages, a message model, for an individual call to the API. And finally, of course, Open Banking needs to make the transactions persistent and for that, we also need a data model and an RDBMS.

The path to achieving our vision is with the FIBO data model. In the first year, we adopt the industry standard, the FIBO Data, and Concept Model. Then we progress to train data architects an OWL. We customize the FIBO to make it our enterprise ontology. The end state then is the full realized Semantic Information Architecture. The years here, they are purely indicative: Some banks may go faster, some banks may go slower. The difference with FIB-DM is that we get an immediate Return on Investment because we can use the FIBO content and design patterns from day one. We don’t have to wait two years until the whole organization has learned RDF/OWL and SPARQL. Also, I believe for Data Architects, it’s easier to learn a new language, OWL, and the tooling if you already know the content and design. You already know the FIBO concepts from the data model. Yeah, and if you want to become a reference implementation, I can teach and advise you on the path to Semantic Enterprise Information Architecture.

About myself, I think I am a recognized FIBO and industry data model expert. If you haven’t done so already, please connect with me on LinkedIn. If you want to look up the profile you can see a long list of recommendations. Here I just want to showcase two of them: The EDM Council, they called me a key participant on the FIBO work. And here Mike Atkin the former director mentioned they had me speak at the industry. These FIBO conference presentations, you can watch the conference videos on the FIB-DM channel. Recently Deutsche Bank Director Shannon Walker said about me – thanks for that. I just want to emphasize, “he trained my architects”.  That is really like teaching them how to use an ontology, it is how to use OWL. You can look all this up on my LinkedIn profile. The FIBO is the authoritative model for financial industry concepts, definitions, and relations. The Enterprise Data Management Council, EDMC is a global association of more than 200 financial institutions. They advocate data management best practices and the development and implementation of data standards. One of these standards is the FIBO, the Financial Industry Business Ontology. Why I say it is superior as a reference model is because EDMC members, banks and investment managers developed the Financial Industry Business Ontology. They developed it as a Business Conceptual Model. and it has more 1,800 normative classes that detail financial instruments, business entities, and processes. The FIBO is both a business model and a schema. In the EDMC’s own words “the Financial Industry Business Ontology is a business conceptual model developed by our members of how all financial instruments business entities and processes work in the financial industry.” So, it’s a business model. Now interestingly when they started the work on FIBO in 2012, the Council choose to specify as a model, not as a conceptual or logical data model, like all these industry data reference models, but they choose to do it in the Resource Description Framework Schema and in Ontology Web Language, in RDF/OWL. They did it because it enables you to specify sample data values, complex constraints, and all that goes beyond the capabilities of a data model. In the data model, all I can do is, I can specify foreign key relationships. The relationships in the model, from one entity to another. In OWL, I can specify constraints across several entities, I can even specify constraints based on data values. That is the more powerful capability of the ontology, and that’s the reason why the Council chooses to release the FIBO in RDF/OWL. However, with that comes a chasm between semantic and conventional data management.

As I said, the FIBO is specified in OWL. However, for OWL we need highly specialized ontologists. The FIBO is very comprehensive but many banks and investment managers don’t have the expertise in-house yet. Very large financial institutions started implementations on RDF Triplestores. However, IT departments must still support and design conventional databases. That is the chasm between the Semantic world and the conventional world. Then, here FIB-DM comes in as a bridge across the chasm, as a bridge for ontology and data architects to work together.

The ontology transformed into a data model leverages the design for relational databases. Again, we have the FIBO in RDF/OWL, and the FIBO deploys on semantic Triples-Stores. One word of caution: I put the store in red. The RDF-Store, the Triple-Store is for knowledge and analytics –  it’s not meant for transaction processing or core banking. And this is where were FIB-DM comes in with the Configurable Ontology to Data model Transformation. I can generate a data model from the ontology and then the data model I can deploy on the relational databases. That leverages the fiber for applications not just in the niche of knowledge management but for the conventional core banking, for our relational world. Atlantic is the way to Semantic Enterprise Information Architecture and Model-Driven Development.

The latest 2020/Q1 release, the full model, now has four thousand five hundred sixty-three entities. It’s the World’s Largest Data Model. Atlantic CODT, that’s where the fish comes in from Configurable Ontology to Data model Transformation, is the technology that created the FIBO data model.

There’s a new release. It’s now shipping two models the Normative and the Informative a data model. Here, this Venn diagram is to depict the two. As a normative model, there is an open-source version.  That’s the data model that you downloaded based on FIBO or 2018/Q4 and it has the modules Foundation, Business Entities, Finance Business and Commerce. and some Upper-Ontology reference models here for languages, countries, and currencies. FIBO and FIB-DM 2020/Q1 Production adds to the Normative models three more packages, Derivatives Indicators, and Securities. Finally, the Informative model based on FIBO 2020 Development, has four additional modules Corporate Actions, Loans, Market Data, Business Processes, and Collective Investment Vehicles, also known as investment funds. This is an ongoing process. The FIBO releases a new version every quarter and a couple of weeks later then the new FIB-DM version gets generated and ships. We said the ongoing process, new models enter FIBO development. They enter here in the informative bubble and some informative modules become normative. So, when a loan is finally signed off it will move up into the normative model.

To show an Open Banking example, we took the European Union, the Revised Payment Service Directive, also known as a PSD2. Here is the standard PSD2 diagram. It shows there are four main actors and their relations. We have the Payment Service User. That can be a person who uses a Third-Party Provider. That’s what’s new about Open Banking third parties, on behalf of the bank customer, can request information, and in the PSD2, can request payments. The payment request goes to an Account Servicing and Payment Service Provider, who in turn validates the requesting user. For this to work, the third-party provider has to be certified by a National Competent Authority. Banks enroll with the NCA to provide the services. You know this standard diagram of the actors. Here, it’s rendered in the FIB Concept Map. In FIBO all these actors are called a Ting in Role. The FIBO Concept Maps uses a mask as a symbol for Thing in Role.

We can expand our initial diagram. We pull in other FIBO concepts from the concept map. All Things in Role have an identity, the Autonomous Agent. In the case of payments here, it is a person and the payment user is identified. It has a photo ID, driver’s license, passport, or something like that. We can also see that an Autonomous Agent may have many roles. The Account Servicing and Payment Service Provider has an identity. It is a bank. In our case, it’s ELB. The Autonomous Agent, the Legal Entity of the bank also can have many roles. ELB may also be a Third-Party Provider, requesting information from other banks.

Altogether, there are 15 Fundamental Concepts in the FIBO and FIB-DM. The concept map, for the benefit of non-technical users, creates mnemonic icons and abbreviations for them. We can see that we draw from the 15 Concepts to further detail our user model. Fundamental Concepts define the Open Banking scope. For example, the provider must be in the Registry that the Competent Authority establishes. That in FIBO or FIB-DM is called an Arrangement. Network locations are locations in the model. Of course, we have products and services. We have customer Agreements, including the agreement to participate in Open Banking. We exchange documents. Agreements are further detailed with Commitments, Contractual Elements, and Legal Constructs. For instance, we would have limits on a credit card on an overdraft. These are commitments by the Financial Institution. Of course, we have a fundamental concept of Account. In Open Banking payments, this is the payment account and the beneficiary account. Finally, our API-Calls, they Occurrences. Another word for that in other models it the event.  That is the same. API-Call is an occurrence. Of course, the occurrence will have at the Timestamp. These 15 Concepts, they are not just a free form. All of these are ultimate supertypes in the data model. Okay, they are main entity hierarchies in FIB-DM. This is how we go from the Concept Map to scope in the data model.

You want to have transparency for your FIB-DM evaluation. You already downloaded the Open-Source Core model, you explored the model in PowerDesigner. You should study the Education Resources and examine the 2020/Q1 model content. You can download a spreadsheet that lists all the 4500 entities and their definitions. Finally look at the upgrade page on the FIB-DM webpage to review license, maintenance, and pricing.

Well, thanks for your interest and enjoy the ride, ELB!  There are two webinars that are recommended for a more detailed treatment of 15 concepts and how to scope of FIB-DM subset model. One is Semantics for Finance Users. It explains all the 15 Fundamental Concepts in detail. The other one is “Scoping our first data model,” which transforms the Concept Map to PowerDesigner and creates a project model. That example is for a Regulatory Report, but you will learn the technique, how to do it for Open Banking as well. That you find on the FIB-DM YouTube channel. The resources here, you find them on FIB-DM.com Just go down the menu with resources and you see the videos as well as the PowerPoints for download. Well, thanks a lot, and have a nice day.